关键词: 过程挖掘算法, Spark集群, 大数据, 并行分布式化

Abstract: Aiming at the problem that the traditional process discovery algorithms had low efficiency for mining models from large-scale event log,a method of using Spark clusters to conduct parallel distributed process mining was proposed.For the process mining algorithm based on the log activity relationship,the method could accelerate the extraction of activity relationship.Through parallel distributed extraction of activity relationships,the event log was transformed into an activity relationship matrix.By using the relation matrix,the process model was mined.The distributed-Mine algorithm and the distributed flexible heuristic miner algorithm were implemented by Spark,and the result showed that the proposed method leaded the current best algorithm implementation in terms of time consumption,and the mining efficiency was improved significantly.

Key words: process mining algorithm, Spark clusters, big data, parallel distribution